Minireview: Cracking the Metabolic Code for Thyroid Hormone Signaling

Abstract

Cells are not passive bystanders in the process of hormonal signaling and instead can actively customize hormonal action. Thyroid hormone gains access to the intracellular environment via membrane transporters, and while diffusing from the plasma membrane to the nucleus, thyroid hormone signaling is modified via the action of the deiodinases. Although the type 2 deiodinase (D2) converts the prohormone T₄ to the biologically active T₃, the type 3 deiodinase (D3) converts it to reverse T₃, an inactive metabolite. D3 also inactivates T₃ to T₂, terminating thyroid hormone action. Therefore, D2 confers cells with the capacity to produce extra amounts of T₃ and thus enhances thyroid hormone signaling. In contrast expression of D3 results in the opposite action. The Dio2 and Dio3 genes undergo transcriptional regulation throughout embryonic development, childhood, and adult life. In addition, the D2 protein is unique in that it can be switched off and on via an ubiquitin regulated mechanism, triggered by catalysis of T₄. Induction of D2 enhances local thyroid hormone signaling and energy expenditure during activation of brown adipose tissue by cold exposure or high-fat diet. On the other hand, induction of D3 in myocardium and brain during ischemia and hypoxia decreases energy expenditure as part of a homeostatic mechanism to slow down cell metabolism in the face of limited O₂ supply.

Most eukaryotic cells are equipped with built-in genetic programs that control cell division, homeostatic functions, and basic phenotype. These cellular programs can be modified by environmental cues, such as nutrient availability or biologically active molecules, the nature and intensity over which most cells have very little control. Hormones constitute an example of such biologically active molecules, and cells are generally classified as responsive or unresponsive, depending on whether they have sufficient number of their cognate receptors. However, it has become increasingly clear that cells are not passive bystanders in this process and instead can actively customize hormonal signals, such as with sexual steroids and thyroid hormone. For example, cellular expression of 5α-reductase or P450 aromatase, respectively, transforms the testosterone molecule into dihydrotestosterone or estradiol, locally changing testosterone's biological activity in opposite directions. A similar scenario exists in the case of the deiodinases, enzymes that can locally activate or inactivate thyroid hormone.

Cellular membranes are relatively impermeable to thyroid hormone and thus membrane transporters are necessary for access to the intracellular environment (1). Once inside the cells, thyroid hormone diffuses toward the nucleus and eventually binds to its receptors, high affinity ligand-dependent transcription factors that modify gene expression (2). However, inside the cells, the prohormone T₄ can be transformed to the biologically active T₃ molecule via the type 2 deiodinase (D2), or it can be inactivated to form reverse T₃ via the type 3 deiodinase (D3). Most importantly, T₃ is also inactivated by D3, preventing or terminating thyroid hormone action (3). Thus, although diffusing from the plasma membrane to the nucleus, thyroid hormone signaling is modified via the action of the deiodinases.

Deiodinases are dimeric integral-membrane thyroredoxin fold-containing selenoproteins of about 60 kDa (dimer) (4–9). Each dimer counterpart consists of a selenocystein-containing globular domain that is anchored to cellular membranes through a single amino-terminal transmembrane segment. D2 is an endoplasmic reticulum (ER)-resident protein that is retained in ER and generates T₃ in the proximity of the nuclear compartment (10). On the other hand, most D3 goes through the Golgi complex and reaches the plasma membrane, where it undergoes endocytosis and recycles via the early endosomes (4). Thus, D2 expression confers cells with the capacity to produce additional amounts of T₃ and thus enhances thyroid hormone signaling. In contrast, expression of D3 results in the opposite action. Furthermore, these events occur in the cell without relative changes to plasma thyroid hormone levels (11, 12).

The Dio2 and Dio3 genes undergo transcriptional regulation throughout embryonic development, childhood, and adult life (11). Dio2 is a highly sensitive cAMP-responsive gene (13) that is also positively regulated by nuclear factor κB (14) and forkhead box (Fox)O3 (15). At the same time, Dio3 is up-regulated by retinoic acid, 12-O-tetradecanoyl phorbol 13-acetate, basic fibroblast growth factor (16), TGFβ (17), the hedgehog-GLI family zinc finger 2 pathway (18), and hypoxia-inducible factor-1α (19).

The D2 protein is unique in that it can be switched off and on via an ubiquitin (Ub)-regulated mechanism, triggered by catalysis of T₄ (20–22). It is assumed that T₄ deiodination exposes Lys-residues in D2's globular domain that are subsequently conjugated to Ub. Moreover, this results in inactivating D2 by disruption of the dimer formation (20). Ub-D2 is not immediately taken up by the proteasome and instead can be deubiquitinated and reactivated to produce another molecule of T₃, repeating the cycle. Although two Ub conjugases are involved in the process of D2 ubiquitination (23, 24), the limiting components of this pathway are two E3-ligase adaptors. These include the hedgehog-inducible suppressor of cytokine signaling-box containing WD repeat suppressor of cytokine signaling box-containing protein-1 (25) and TEB4 (26), a ligase involved in the ER-associated degradation program. In contrast, two Ub-specific proteases (USP), USP20 and USP33, mediate deubiquitination and reactivation of Ub-D2 (21).

Thus, it is clear that thyroid hormone levels in the plasma do not faithfully reflect thyroid hormone signaling in cells; this action takes place inside the cell. A complex network of transcriptional and posttranscriptional mechanisms regulating deiodinase expression is at work in health and disease, mediating rapid customization of thyroid hormone signaling on a cell-specific basis.

Deiodinases and the Metabolic Effects of Thyroid Hormone

Insulin typifies how metabolic pathways are controlled by systemic hormones. In the minutes that follow a meal, insulin is secreted into the bloodstream, exposing all tissues to elevated levels of insulin. As a result, glucose uptake and oxidation is increased, the synthesis of fatty acid is accelerated, and protein anabolism is enhanced. A few hours later, plasma insulin levels are back to premeal levels and so are its metabolic effects. This is a very different model to that which cells respond to thyroid hormone. On the contrary, thyroid hormone levels in the plasma hardly fluctuate in healthy individuals, as shown in a year-long study of serum levels of T₄ and T₃ (27). Thus, thyroid hormone-responsive metabolic processes are turned on and off by thyroid hormone via deiodination pathways that are taking place inside the target cells, seemingly invisible from the plasma viewpoint (11).

Even though we are just starting to understand these pathways, some thyroid hormone effects on metabolism are well recognized, including acceleration of substrate cycles, ionic cycles, and mitochondrial respiration, all leading to accelerated energy expenditure (28). Unfortunately, most of what we know about these effects is from nonphysiological models in which subjects were systemically hypothyroid or thyrotoxic. To illustrate this point, hypothyroidism and thyrotoxicosis are known to affect sympathetic outflow to a number of metabolically active tissues, such as white and brown adipose tissue (BAT), liver, skeletal muscle, and heart (29, 30). These states mask the true effects of thyroid hormone deficiency or its excess. This is illustrated in the D2 knockout (D2KO) mice. At room temperature, which is considered a significant thermal stress for mice, D2KO mice preferentially oxidize fat, have a normal sensitivity to diet-induced obesity, and are supertolerant to glucose load. However, when thermal stress is eliminated and sympathetic activity minimized at thermoneutrality (30 C), an opposite phenotype is encountered, one that includes obesity, glucose intolerance, and exacerbated hepatic steatosis (31). Thus, the cell-specific metabolic effects of thyroid hormone are largely unknown, and cracking the code requires understanding the deiodinase pathways.

A glimpse into this world is available through the studies in which D2 and D3 expression reciprocally affect energy expenditure in a number of cell and animal models. For example, cAMP-dependent induction of D2 expression during activation of brown adipocytes by cold exposure or high-fat diet enhances local thyroid hormone signaling and energy expenditure, the absence of which prevents normal BAT function (31–34). On the other hand, hypoxia-inducible factor-1α-dependent induction of D3 in myocardium and brain during ischemia and hypoxia decreases energy expenditure, supposedly as part of a homeostatic mechanism to slow down cell metabolism in the face of limited O₂ supply (19, 35). In fact, D3 reactivation in disease states can be so powerful that it compromises systemic thyroid economy, leading to euthyroid sick syndrome (36). In rare instances, D3-mediated thyroid hormone inactivation is so dramatic that it exceeds the thyroidal synthetic capacity to sustain thyroid economy, leading to consumptive hypothyroidism (37).

D2 expression is the target of a rapidly growing number of molecules that accelerate energy expenditure and metabolic programs in cells and animal models. These include bile acids (38), flavonols (39), and chemical chaperones (40), which in-turn confer protection against diet-induced obesity. Insulin and peroxisome proliferator-activated receptor γ agonists are also bona fide inducers of D2 in skeletal muscle (41). On the other hand, signaling through the D2 pathway is dampened by ER stress (42) and the LXR-RXR pathway (43), the metabolic consequence of which is currently under investigation.

Deiodinases and the Development of Metabolically Relevant Tissues

During vertebrate embryogenesis, developmental signals control the expression interplay between D2 and D3 in metabolic relevant tissues, such as BAT (44), pancreatic islets, and skeletal muscle (15), explaining how “systemic” thyroid hormone can affect local control of tissue embryogenesis.

In the 3-d developmental snapshot during which BAT develops in mice [embryonic day (E)16.5–E18.5], D2 expression is up-regulated about 5-fold and D3 expression drops by 75%. This results in increased local net T₃ availability, whereas serum T₃ remains unchanged. This rapid enhancement in thyroid hormone signaling is critical for the expression of genes defining BAT identity, i.e. uncoupling protein (UCP)1, peroxisome proliferator-activated receptor gamma co-activator-1α, and Dio2 (44). Notably, these changes in gene expression are observed in utero, without a thermogenic challenge, which highlights the relevance of D2 and its ability to amplify thyroid hormone signaling in a developmental setting. The inactivation of the Dio2 gene as in the D2KO mouse results in a permanent BAT thermogenic defect, compromising thermoregulation and the ability to dissipate excessive calories from diet (31, 32).

The D2 pathway seems to also be critical for skeletal muscle development and function (15). Besides regulating insulin sensitivity in myocytes (41), D2-mediated T₃ production is also required for the T₃-dependent expression of myogenic factors, such as the myogenic regulatory factor (MyoD), which drive myocyte development. As in BAT, myocytes from D2KO mice have impaired development and function supposedly due to lower intracellular T₃ generation. The control of the D2 pathway in myocytes is dependent on the transcription factor FoxO3, which directly binds to the Dio2 promoter, up-regulating D2 expression. The fact that FoxO3 KO myocytes also display impaired cellular development, easily reversed by the addition of exogenous T₃, underscores the physiological relevance of the FoxO3/D2 interplay.

The opposite scenario is observed during development of the pancreatic β-cells, with D3 expression keeping thyroid hormone signaling to a minimum, from late embryonic development throughout adulthood (45). The late emergence of D3 expression at E17.5 is restricted to insulin positive cells, indicating a focused role in β-cell but not α-cell development. α-Cell development occurs at a much earlier phase of embryogenesis (by E9.5). As a result of untimely expression of thyroid hormone, D3KO animals exhibit a reduction in total islet area due to decreased β-cells area, insulin content and lower expression of key islet genes involved in glucose sensing, insulin expression, and exocytosis. This is physiologically significant given that adult D3KO animals are glucose intolerant due to impaired glucose-stimulated insulin secretion, without changes in peripheral sensitivity to insulin.

Deiodinases in the Medial Basal Hypothalamus (MBH)

In the central nervous system, D2 is expressed in astrocytes, whereas thyroid hormone receptor and D3 are found in adjacent neurons (46, 47). Thus, glial cell D2 produces T₃, which acts in a paracrine fashion to induce thyroid hormone-responsive genes in the nearby neurons (35), a process that is also modulated by D3 activity in the neurons. This paracrine pathway of thyroid hormone action depends on the deiodinases and is thus regulated by signals such as hypoxia, hedgehog signaling, and lipopolysaccharide-induced inflammation, as evidenced both in vitro as well as in rat models of brain ischemia and mouse models of inflammation (48). Therefore, as in other tissues, it is clear that deiodinases function as control points for the regulation of thyroid signaling in the brain.

The neurons in MBH are a target of thyroid hormone, and thus, local D2 and D3 expression can affect thyroid economy and a number of other homeostatic functions (46, 47, 49, 50). Within the MBH, D2 expression is largely restricted to the tanycytes, which are ependymal cells lining the floor and infralateral walls of the third ventricle extending from the rostral tip of the median eminence (ME) to the infundibular recess, surrounding blood vessels in the arcuate nucleus (ARC), and in the ME adjacent to the portal vessels and overlying the tuberoinfundibular sulci (49, 50). Thus, the tanycytes seem to be a major source of T₃ to the ARC-ME region of the hypothalamus, likely with important metabolic consequences.

For example, hypothalamic D2 activity in rodents exhibits a circadian rhythmicity with an activity peak at night, which coincides with their peak of metabolic activity (51). At the same time, fasting induces a state of central hypothyroidism that has been linked to an approximately 2-fold up-regulation of D2 expression in the hypothalamus (52) and suppression in TRH/TSH secretion. It has also been suggested that D2 expression in the ARC is localized in glial cells that are in direct opposition to neurons coexpressing neuropeptide Y, Agouti-related peptide, and UCP2 (53). Notably, the fasting-induced increase in D2 activity and local thyroid hormone activation in the ARC is paralleled by an increase in UCP2-dependent mitochondrial uncoupling in neuropeptide Y/Agouti-related peptide expressing neurons. These events were shown to be linked to the increased excitability of these orexigenic neurons and consequent rebound feeding after food deprivation (54).

Deiodinases and the Skeleton

There are strong links emerging between metabolic control and bone and the role of the deiodinases in the skeleton (55). The skeleton is a target of thyroid hormone, which responds by accelerating bone turnover to the extent that there is a net loss of bone mass during systemic thyrotoxicosis (56). In mice, thyroid hormone signaling is kept to a minimum during early bone development due to the high D3 expression (57). Later, during E14.5–E18.5, there is a decrease in D3 and an increase in D2 expression, thus increasing thyroid hormone signaling toward the end of gestation (57). Studies in the developing chicken skeleton indicate that hedgehog signaling mediates the reciprocal control of D2 and D3 expression, transcriptionally increasing Dio3 gene expression and inactivating D2 via induction of WSB-1, the E3 ligase Ub adaptor that ubiquitinates D2 (18, 25). In adult mice, D2 is present in whole-bone extracts, as well as in skeletal cells and differentiated osteoblasts (58), but it is undetectable in chondrocytes and osteoclasts (59). Its absence, as in the D2KO mouse, results in brittle bones due to reduced bone formation, without changes in bone resorption (60). T₃ target gene analysis indicates osteoblastic T₃ deficiency (60), suggesting that D2-mediated T₃ production in osteoblasts is important for maintenance of adult bone mineralization and optimal bone strength.

Conclusion

Thyroid hormone signaling is a local event, with target cells playing a major role through controlled expression of the activating or inactivating deiodinases. Although it is conceivable that plasma T₃ plays a metabolic role in some tissues, its relative constancy throughout adult life precludes it from controlling major metabolic pathways. The local role played by the deiodinases in customizing thyroid hormone signaling is the predominant modus operandi through which thyroid hormone exerts its metabolic effects, including in the BAT, β-cell, MBH, bone, and skeletal muscle. Much of this new paradigm of thyroid hormone action was validated through the study of mice with targeted disruption of the deiodinase genes (61). Much remains to be learned while we decipher this code, particularly through the use of a new generation of tissue-specific deiodinase KO animals (62). The consequence of this new way of looking at thyroid hormone action has very significant clinical implications, because serum hormone levels may not be predictive of events that are driving clinical symptoms. This is well illustrated by the series of reports correlating polymorphisms in the three deiodinase genes with a growing number of diseases and clinical conditions in individuals with normal thyroid function tests (63).